You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flume.apache.org by "Edward Sargisson (JIRA)" <ji...@apache.org> on 2014/07/03 20:12:34 UTC

[jira] [Commented] (FLUME-1793) Unit test TestElasticSearchLogStashEventSerializer fails with IBM JDK

    [ https://issues.apache.org/jira/browse/FLUME-1793?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14051775#comment-14051775 ] 

Edward Sargisson commented on FLUME-1793:
-----------------------------------------

Hi, just catching up on my emails now.

Thanks for raising this and adding the patch. It's appreciated.

IMO, if we have an ordering problem then the assertions should be written to ignore ordering - and not be conditional on the JVM vendor. There's nothing stopping this breaking even in the Oracle JVM.

If you felt like making that change that would be great. Once it's sorted, please put it into a review in the Apache Review Board.

> Unit test TestElasticSearchLogStashEventSerializer fails with IBM JDK
> ---------------------------------------------------------------------
>
>                 Key: FLUME-1793
>                 URL: https://issues.apache.org/jira/browse/FLUME-1793
>             Project: Flume
>          Issue Type: Bug
>          Components: Sinks+Sources, Test
>    Affects Versions: v1.3.0
>         Environment: IBM Java 6 SR 12
> RHEL 6.3
>            Reporter: Aline Guedes Pinto
>            Priority: Minor
>              Labels: test-patch
>         Attachments: FLUME-1793.patch
>
>
> TestElasticSearchLogStashEventSerializer fails with IBM JDK:
> testRoundTrip(org.apache.flume.sink.elasticsearch.TestElasticSearchLogStashEventSerializer)  Time elapsed: 0.201 sec  <<< FAILURE!
> org.junit.ComparisonFailure:expected:<.....@...> but was:<.....@...>
> The values on the "@fields" element are the same as expected, but the order is different. This test case works with SUN Java though - it seems like HashMap on IBM Java and SUN Java return values in different orders, and the test case is counting on order. Order shouldn't matter when using a hashmap though.



--
This message was sent by Atlassian JIRA
(v6.2#6252)